A Newspaper's 900 Telephone Poll: Its Perceived Credibility and Accuracy

Abstract

Comparing results of a telephone poll based on a random sample with those from a call-in poll (at 50 cents per call for unlimited calls) shows that call-in polls contain very substantial errors. Even so, as many as one-third of newspaper readers held call-in polls in high regard, especially readers with less formal education. This study suggests call-in polls cash in on the respect gained in recent years by soundly conducted polls, and that newspapers should avoid call-in polls or at least clearly warn readers that results may be widely at variance with real opinion.